📚 AI's literacy cliff
1. Key Themes
AI Is Masking a Structural Workforce Skill Deficit
AI tools are enabling workers to complete tasks they don't fully understand, concealing a pre-existing but massive literacy crisis. This creates systemic fragility — the workforce appears productive until higher-order judgment is required.
"The net effect of AI on the workplace is probably going to be increased demand and need for workers with higher levels of basic skills, not lower." — Stephen Reder, Portland State University
"Not only are skill levels going down, but particularly among people at the lower end of the skill spectrum, the amount that they use the skills that they have is going way down." — Stephen Reder
The Scale of the Literacy Problem Is Larger Than Most Realize
The underlying data reveals a crisis that dwarfs most public discourse about it: 130 million U.S. adults read below a sixth-grade level, and 43 million cannot read, write, or do basic math above a third-grade level (ProLiteracy). Yet this rarely features prominently in workforce policy debates.
"If you can't read, write, speak the language, can't use a computer, your chances of being gainfully employed are pretty slim." — Sharon Bonney, CEO, Coalition on Adult Basic Education
"More than 90% of jobs require some form of computer literacy." — Sharon Bonney
AI Training Data Reflects and Amplifies Western Cultural Dominance
Most large language models are built on data that originates predominantly from North America and Europe, systematically underrepresenting non-Western populations and producing errors that persist even after diversity-focused remediation efforts.
"Most mainstream models are trained on the work of Western writers — particularly white men — and regularly mimic those values, writing styles, viewpoints and biases."
"Those training materials have resulted in LLMs inventing details based on Western assumptions about cultural traditions or values, and those errors persist despite Big Tech putting in work to train them with more diverse viewpoints and data."
Government Equity Participation in AI and Tech Companies Is Becoming a Policy Pattern
The Trump administration has moved from passive regulator to active equity stakeholder in AI-adjacent companies, acquiring shares in chipmakers, miners, and quantum computing firms — a structural shift with long-term implications for how tech companies access federal capital.
"The U.S. government now owns shares of chipmakers, miners and quantum computing companies, often taking a stake in exchange for federal money that was originally meant to come with no ownership strings attached."
Trump "said Friday that he 'should be a stockbroker,' pointing to his administration's recent deal to take a stake in Intel."
2. Contrarian Perspectives
AI Raises — Not Lowers — the Baseline Skills Required to Work Effectively
The dominant narrative is that AI democratizes work by lowering the skill floor. The article argues the opposite: AI-assisted outputs still require workers to evaluate, verify, and exercise judgment — meaning the bar for functional competency may actually be rising.
"The net effect of AI on the workplace is probably going to be increased demand and need for workers with higher levels of basic skills, not lower." — Stephen Reder
Supporting evidence: Researchers describe "cognitive surrender" — deferring to AI outputs without evaluating them — as a distinct risk. The calculator analogy is apt: calculators didn't eliminate the need for mathematical understanding; they raised the cost of not having it.
Book Sales Are Not a Proxy for Literacy — And May Be Misleading Investors and Policymakers
The recovery of Barnes & Noble and growth of independent bookstores might suggest reading culture is healthy. The article explicitly rebuts this.
"However, Reder said book buying and literacy skills are not the same thing. The bigger divide may be between people who use reading deeply in everyday life and those who rarely practice those skills."
This is significant for anyone using consumer book market health as a signal of educational or cognitive workforce capacity.
Data Extraction From Marginalized Communities Is a Form of Ongoing Colonialism, Not a Historical Artifact
The conventional view treats colonialism as a closed historical chapter. Critics cited in the article argue AI's data practices constitute a continuation of the same dynamic under a new mechanism.
"Colonialism is always portrayed as something that happened in the past... many countries got independence, and then the textbooks say 'colonialism is over.'" — Julian Posada, Yale
"To say, 'Well, it's just out there. We can just take it.' That was what colonialism was about, just taking everything." — Nick Couldry, co-author, Data Grab
3. Companies Identified
Barnes & Noble
- Description: National brick-and-mortar book retailer
- Why Mentioned: Used as a data point to challenge the assumption that book market recovery signals broad literacy health
- Quote: "Barnes & Noble has staged a comeback, suggesting reading culture remains strong for some."
- Description: Major U.S. semiconductor manufacturer
- Why Mentioned: Cited as a specific example of the Trump administration taking an equity stake in a tech company
- Quote: Trump pointed "to his administration's recent deal to take a stake in Intel."
- Description: AI company behind ChatGPT
- Why Mentioned: Reported to be revamping ChatGPT to appeal more to enterprise customers (brief mention, sourced from Financial Times)
- Quote: "OpenAI plans to revamp ChatGPT to make it more appealing to enterprise customers, according to sources."
4. People Identified
- Description: Professor Emeritus of Applied Linguistics, Portland State University
- Why Mentioned: Primary academic voice on AI's relationship to workforce literacy; argues AI raises rather than lowers baseline skill needs
- Quote: "The net effect of AI on the workplace is probably going to be increased demand and need for workers with higher levels of basic skills, not lower."
- Description: CEO, Coalition on Adult Basic Education
- Why Mentioned: Provides ground-level context on what low literacy looks like in workforce pipelines and apprenticeship programs
- Quote: "If you can't read, write, speak the language, can't use a computer, your chances of being gainfully employed are pretty slim."
- Description: Senior Fellow, National Skills Coalition
- Why Mentioned: Coined the concept of AI creating an "invisible drag on productivity" and highlighted how low literacy among supervisors can cascade across entire organizations
- Quote: "If it's flashing a red warning light that says we have a literacy challenge, then we probably really do have a literacy challenge."
- Description: Assistant Professor, Yale University; studies human labor and data production
- Why Mentioned: Central voice in the AI-as-colonialism argument, connecting current data extraction practices to ongoing colonial dynamics
- Quote: "Colonialism is always portrayed as something that happened in the past... many countries got independence, and then the textbooks say 'colonialism is over.'"
Aditya Vashistha
- Description: Assistant Professor, Cornell University
- Why Mentioned: Provides a concrete example of how Western-biased training data produces flattened, inaccurate representations of non-Western cultures
- Quote: "You will find different regional cuisines which differ in the spices which are used, or in what moderation, like the amounts they use."
Nick Couldry
- Description: Co-author, Data Grab: The New Colonialism of Big Tech and How to Fight Back
- Why Mentioned: Frames AI data collection as structurally analogous to imperial land seizures
- Quote: "To say, 'Well, it's just out there. We can just take it.' That was what colonialism was about, just taking everything."
Sriram Krishnan
- Description: White House AI Adviser
- Why Mentioned: Departing his role at the end of June — a notable shift in the administration's AI policy infrastructure
- Quote: "Sriram Krishnan plans to leave his role as a White House AI adviser at the end of the month."
5. Operating Insights
Don't Mistake AI-Assisted Output for Verified Competence — Audit the Underlying Skills
For operators deploying AI tools in workforces with mixed skill levels, AI-generated productivity can mask skill gaps that only surface under pressure. The risk is systemic: low-literacy supervisors can impair compliance and performance across entire teams.
"Low literacy among supervisors can ripple across entire workplaces, affecting performance and compliance." — Amanda Bergson-Shilcock
Implication: Build evaluation checkpoints that test comprehension and judgment, not just output volume, especially in roles involving safety, compliance, or customer-facing decisions.
Workers with Higher Foundational Skills Will Be More Valuable, Not Less, as AI Scales
Contrary to fears of AI replacing skilled workers, the article suggests employers should prioritize hiring and upskilling workers who can critically evaluate AI outputs rather than those who simply use the tools.
"You still need to know what you're doing." — Stephen Reder
Implication: Literacy, critical thinking, and digital skills become a competitive differentiator in AI-augmented hiring — worth investing in through training programs, apprenticeship pipelines, or targeted upskilling partnerships.
6. Overlooked Insights
Consent and Accuracy Gaps in Data Collection From Marginalized Groups Represent a Latent Legal and Reputational Risk
The article notes that data collection from Indigenous groups and people of color is "often done without their consent or any verification that the information is accurate." As regulatory scrutiny of AI training data grows globally, this represents an underappreciated liability for companies building or licensing LLMs — and a potential opportunity for platforms that build verified, consent-based diverse data pipelines.
"Data collection from these groups is often done without their consent or any verification that the information is accurate."
The "Invisible Drag" Problem Has No Current Measurement Mechanism
Bergson-Shilcock explicitly notes the literacy-driven productivity drain "doesn't show up in data." This is both a risk signal for operators and a market gap — there is no established tool or methodology for quantifying how hidden literacy gaps suppress team performance in AI-augmented environments.
"An 'invisible drag on productivity' that doesn't show up in data but slows teams down." — Amanda Bergson-Shilcock